Transcript The Parallel Climate Model - Computational Information Systems

The Parallel Climate Model (PCM)
and Transition to a Climate
Change Version of the
Community Climate System
Model (CCSM)
Warren M. Washington
NCAR
Climate Change Simulations
 Present
and future climate change and
assessment simulations
 Merged
 Future
CSM and PCM model
research needs
 Present
and Future cooperation between
NSF and DOE
Partial List of Distributed Involvement
DOE and NSF Supported Project with:











Los Alamos National Laboratory
National Center for Atmospheric Research
Naval Postgraduate School
Oak Ridge National Laboratory
University of Texas, Austin
Scripps Oceanographic Institute
DOE Program on Climate Diagnostics and Intercomparison
U.S. Army Cold Regions Research and Engineering
Laboratory
National Energy Research Supercomputer Center
Lawrence Berkeley National Laboratory
Others
Contributions to the DOE Climate
Change Prediction Program (CCPP)
 Develop
climate modeling capability that takes
advantage of new generation parallel architecture
supercomputers
 Build
on the previous DOE CHAMMP modeling
developments
 Develop
model components and coupled models
that can be used for energy policy, IPCC, and
future National Assessments
History
CSM1 and PCM1
 Built
for vector/parallel
Computers system
 Atmosphere: CCM3
 Ocean component:
NCAR ocean model
 Sea ice dynamics of
Flato-Hibler dynamics
and simplified
thermodynamics




Built for distributed
parallel Computer system
Atmosphere: CCM3
Ocean component: LANL
Parallel Ocean Program
(POP)
Sea ice Model -Naval
Postgraduate School
model: viscous plastic
dynamics; simplified
thermodynamics
Coupler Overview
• Enables coupled model to be
constructed from independent
components
• Controls execution of
component models
Land
• Computes interfacial fluxes
• Implements conservative
mapping of fluxes of
momentum, energy, and water
between different grids
Atmos
Coupler
Ocean
Ice
Coupler: Design Goals
• Support both sequential & distributed execution
• High-performance scalable parallel
implementation of performance-sensitive
regridding and communication operations
– Address parallelism and load balance issues
• Run-time rather than compile-time specification
of processor allocations
• Support for stand alone execution
Merging of CSM and PCM
 Agreement
to use the same model components
 CSM,
PCM, and DOE lab staff will develop a
merged flux coupler b– ongoing SciDAC team of
NCAR and DOE laboratory involvement
 Full
merger occurs when the CCSM is available for
climate change simulations and the new flux
coupler
 NSF
and DOE efforts may use different resolutions
with DOE emphasis on high resolution studies for
regional climate change studies
Computational Design Question
Parallel Integration
ATMOSPHERE
OCEAN
SEA ICE
RIVER
TRANSPORT
COUPLER
LAND/
VEGETATION
ATMOSPHERE
Sequential Integration
(stacked)
COUPLER
OCEAN
SEA ICE
RIVER
TRANSPORT
LAND/
VEGETATION
T
I
M
E
T
I
M
E
Coupled System Processor Layout
CCSM2
PCM
CSM1
Target?
PCM Version 1 - NERSC Machines
Simulated Years per Wallclock Month
250
64pes
IBM SP
NHII
200
Simulated Years
IBM SP
WHII
150
IBM SP
WHI
T3E-900
100
T3E-600
50
T3D
0
1995
10/24/2001
1996
1997
1998
Year
1999
2000
2001
David Pierce
Scripps
Institution of
Oceanography
Examples of Climate
Change Experiments

Greenhouse gases

Sulfate aerosols (direct effect)

Stratospheric ozone

Land surface changes

Volcanic forcing

Solar change forcing

Carbon soot aerosol (Indian-Asian region)

Various energy/emissions use strategies
PCM Historical and Future
Simulations

Use of CSM greenhouse gas and sulfate aerosol forcing

1870 control simulation

Historical 1870 to present

IPCC “Business as Usual” assumption

IPCC stabilization assumption

Ensemble of 10 for Historical, BAU/STAB ensemble 5

Solar variability simulation-ensemble of 4

Simulations to year 2200-ensemble BAU/STAB 3
PCM and CSM Presence in the International
Climate Modeling Community
Both prominent in the IPCC Third Assessment Report (2001)
Both represented in the IPCC Data Distribution Centre
(Hamburg)
Both represented in the CLIVAR Coupled Model
Intercomparison Project (CMIP): CMIP1, CMIP2, CMIP2+
Access to CSM: via NCAR (CSM web page)
Access to PCM: runs archived at PCMDI (contact Mike
Wehner: [email protected])
Interim Model:
“PCM-CSM Transitional Model” (PCTM)
 POP
with GM and KPP (LANL, NCAR, NPS), R.
Smith grid modifications (LANL)
 C.
Bitz sea ice multi-thickness(5) distribution
thermodynamics and E. Hunke et al. elastic
viscous plastic dynamics (U. of Washington,
LANL, NCAR)
 River
Transport, Branstetter and Famiglietti (U. Of
Texas, Austin, NCAR)
Future Developments
and Simulations

More simulations with Black Carbon distributions in PCM1
over entire globe

Higher resolution atmosphere -T85d simulations

Improved archival and cataloging of large data sets EARTHGRID/DOE/

Simulations related to DOE mission on energy use
impacts on the climate system - ACPI demonstration
project

Work with statistical community on signal to noise
 CCSM2
climate change simulations early
next year
Why T85?
 Improved
regional simulation of climate change.
 Possible
improvement of the winds in the Arctic
which influence strongly the sea ice distribution.
 Improvement
of orographic precipitation
distributions.
 Improved
to coasts.
simulations of the ocean/sea ice close
I have a Dream!
Remember
the famous
Martin Luther King
Speech!
Organization of
Community Climate System Model
CCSM SSC
Atmosphere
Model
Working Group
Paleoclimate
Working Group
Ocean Model
Working Group
Land Model
Working Group
CCSM
Advisory Board
Polar Climate
Working Group
Climate Change
Software
Climate Variability Biogeochemistry
and Assessment
Engineering
Working Group
Working Group
Working Group Working Group
Future Goals for CCSM
 Improve
Existing Climate Simulation
 Expand
System to Include:
- Biogeochemical Processes
- Atmospheric Chemical Processes
- Ice Sheet Dynamics
- …
 Expand
Outreach and Visibility
 Increase
Computational Resources
The End